Emotion Recognition



Emotion Recognition (ER) service recognizes emotions based on the user's expressions, voice, and typed inputs.

The ER service's recognition of human emotions enables natural verbal and nonverbal communication and allows the human to interact with a machine.


Process of ER Engine


ER service by LG AI Platform supports the following features:

ER Engine Feature
Feature Description
Various input data

Uses audio, image, and text as input data

Various emotion recognition

Determines seven emotions (anger, disgust, fear, happiness, sadness, surprise, neutral).


Recognizes the user's emotions in two ways.

  • Unimodal: Uses a single modality
  • Multimodal: Uses multiple modalities simultaneously
Server Communication

Connects to the server using the HTTP/2 method based on transport layer security (TLS) for enhanced security.



All functions of ER service are run on a server. ER service takes text, visual, and audio data as input and generates appropriate responses.


ER Engine Architecture


Examples of Use

ER service is used a lot in everyday life.


  • Friendship with AI robots

Allows humans and robots to interact and communicate with each other emotionally.


AI robots act like friends

  • Chatbot service that recognizes users' emotions

Allows customer issues to be escalated quickly through real-time monitoring of the chatbot service. Through recognizing customer emotions and complaints, businesses can take action so that agents, not the chatbot, can respond.

Chatbot service that recognizes users' emotions